CoreWeave

CoreWeave's Mercado Libre case study describes a three-week transition from proof of concept to production and migration of query-expansion inference workloads to CoreWeave Kubernetes Service without disruption.

CoreWeave expanded major AI infrastructure agreements in 2026, including a $21B expanded agreement with Meta, a multi-year agreement with Anthropic, and production workloads across financial services, agents, commerce and research.

CoreWeave expanded major AI infrastructure agreements in 2026, including a $21B expanded agreement with Meta, a multi-year agreement with Anthropic, and production workloads across financial services, agents, commerce and research.

CoreWeave's SUNK engineering publication details a custom Nodeset resource, a Syncer reconciling Slurm and Kubernetes state, and a scheduler placing Kubernetes workloads on Slurm-allocated nodes. These concrete shared-pool scheduling mechanisms are independent of its prior MLPerf, ARIA and general optimization evidence.

CoreWeave set MLPerf Training v6.0 records, launched ARIA for autonomous research iteration, and continues full-stack optimization across training, inference, observability and reinforcement learning.

CoreWeave set MLPerf Training v6.0 records, launched ARIA for autonomous research iteration, and continues full-stack optimization across training, inference, observability and reinforcement learning.

CoreWeave reported fiscal 2025 revenue of $5.131 billion and a GAAP net loss of $1.167 billion in its February 26, 2026 results. These realized financial measures are distinct from generic capacity, financing and customer-contract announcements; revenue growth alone is not treated as profitability.

CoreWeave is operating a rapidly expanding global AI cloud, adding APAC capacity, financing large infrastructure build-outs, and serving hyperscale AI customers with multi-billion-dollar agreements.

CoreWeave is operating a rapidly expanding global AI cloud, adding APAC capacity, financing large infrastructure build-outs, and serving hyperscale AI customers with multi-billion-dollar agreements.

CoreWeave's SUNK technical description combines Slurm scheduling with Kubernetes containerized deployments and GitOps, allowing Kubernetes and Slurm workloads to use a shared compute pool.

CoreWeave work spans GPU infrastructure, distributed systems, model training, inference, RL, observability, physical AI and enterprise deployment, creating highly portable systems-AI skills.

CoreWeave work spans GPU infrastructure, distributed systems, model training, inference, RL, observability, physical AI and enterprise deployment, creating highly portable systems-AI skills.